Replication of charsets and timezones should shortly be changed to not need ONE_SHOT
(but 5.0 should still be able to read it, to be able to replicate 4.1 masters).


mysql-test/r/rpl_charset.result:
  result update after the merge (only binlog positions differ from 4.1)
mysql-test/r/rpl_timezone.result:
  result update after the merge (only binlog positions differ from 4.1)
sql/log.cc:
  use ONE_SHOT to save charset and TZ info in the binlog (until I rewrite this soon)
sql/set_var.cc:
  Even if in the future, 5.0 will not use ONE_SHOT in ITS binlog, it must still be able to interpret it, to make 4.1->5.0 replication possible
sql/set_var.h:
  Even if in the future, 5.0 will not use ONE_SHOT in ITS binlog, it must still be able to interpret it, to make 4.1->5.0 replication possible